Developing multi-scale models of bimetallic catalysts for the hydrodeoxygenation of bio-oil compounds

With the ever-increasing need to find a sustainable source of renewable energy, bio-oil produced via the fast pyrolysis of biomass is a promising source of liquid fuel; however, the resulting bio-oil contains oxygenated products that contribute to poor fuel quality [1]. Hydrodeoxygenation (HDO) is used to refine the bio-oil by reducing the oxygen content, ideally using a minimal amount of H₃ in order to form H₃O. Bimetallic catalysts such as Pd on Fe have demonstrated synergistic behavior that contributes to a more cost-effective and longer-lasting catalyst [2-4]. Similarly, Pt/Sn catalysts have also exhibited traits favorable for the HDO of phenolic compounds, namely a propensity to promote lower temperature desorption of aromatic compounds like benzene [5]. Identifying the cause of the different synergetic effects of these systems can provide insight into the synthesis of a superior HDO bimetallic catalyst. However, the behavior of bimetallic catalysts is complex and necessitates a thorough understanding of the nanoscale behavior of these systems in order to develop a truly predictive model on relevant time scales. We present two DFT studies examining first, the interactions between a surface and its adsorbates and second, the interactions between vicinal adsorbates on a surface. While the addition of Pd to Fe catalysts exhibits a synergistic interaction [2-4], further investigation is needed in order to conclude with greater certainty the role of Pd. We have developed a preliminary microkinetic model of the dissociation of H₃O on Fe (100) and on two surface alloys with different concentrations of Pd. We found that increasing the surface concentration of Pd from 0% to 50% caused a ~0.7 eV decrease in the adsorption strength of O, while only a ~0.1 eV decrease in the adsorption of H. This trend is consistent with other studies that investigated the influence of noble metal dopants on the adsorption of benzene and phenol [6, 7]. Moreover, after determining the activation energies for each reaction step from linear scaling relationships [8], it was found that Pd increases thebarriers for both the dissociation of H₃O in OH and H and the subsequent dissociation of OH. These results show that the primary function of Pd is the disruption of the formation of an oxide, due to the aversion O demonstrates for Pd in the Pd/Fe surfaces. Consequently, the risk of catalyst deactivation due to the formation of a surface oxide decreases with the addition of a Pd dopant on an Fe surface, thereby promoting the HDO with minimal deactivation of the active Fe surface. Quantifying the lateral interactions between adspecies is also necessary in the development of truly predictive, theoretical models for bimetallic catalysts as they will significantly affect the coverage and diffusion of adspecies and therefore the heterogeneous catalytic environment. However, such a characterization of adspecies' lateral interactions for the highly complex HDO reaction has not yet been attempted. To such end, we have characterized the benzenebenzene lateral interactions on Pt (111) and PtSn (111) surfaces. These systems have been well-characterized experimentally using temperature-programmed desorption (TPD) [5]. Here, we studied the effect of surface coverage on the adsorption energy of benzene on Pt (111) and Pt₃Sn (111) at different adsorption sites by modeling the TPD of benzene. We found that a mean field model is sufficient in describing the interactions between vicinal benzene molecules (Figure 1). And while it was previously speculated that the broad desorption peak of benzene on Pt (111) was the result of desorption from two adsorption sites of different binding strengths [5], our results indicate that the behavior is more likely due to coverage effects on the surface. Furthermore, we found that benzene's adsorption was significantly weakened on the Pt₃Sn (111) as compared to Pt (111), which is consistent with experimental TPD [5]. Our investigation of modeling the lateral interactions of an aromatic molecule on both the Pt (111) and the Pt₃Sn (111) surfaces justifies future application of our method onto systems less experimentally characterized, such as doped Fe surfaces.